Parasolid, the CAD kernel owned by Siemens, is more than 30 years old at this point. The same amount of effort has gone into many of the other building blocks as well. OPENCASCADE is a FOSS CAD kernel but doesn't work well for real CAD, and support is hard to come by for companies building on it.
Most startups tend to license these tools instead of building them, but that comes at a large cost that must be pushed down to customers, hence the $$$$ cost of CAD today.
Plus, they are not shy about acquisitions. It's how they maintain their dominance. Any company that builds a remotely useful product in that space will find themselves sitting in front of a Big Fucking Check that's going to be hard to turn down. Anyone who turns it down will probably find themselves sitting in front of several patent lawsuits all over the world.
Answer: It's extremely hard to build any equivalent to Fusion 360 or any full featured CAD/CAM tool. This is not a weekend hackathon problem!
There are so many fundamental building blocks, with many requiring teams of PhDs to produce: CAD kernels, constraint solvers, file format interop, sketching tools, assembly support (mating), simulation, etc... Not to mention the manufacturing specific modeling tools for sheet metal, generative design for milling & 3D printing.